Real-Time Rolling Stock Rescheduling with Dynamic and Uncertain Disruption Information & Integration of Rolling Stock Rescheduling and Shunting Operations

In the first half, we consider rolling stock rescheduling in a setting where disruption information becomes available dynamically and is incomplete and often incorrect. In particular, we consider disruptions for which the duration and severity are not immediately known and can change as the disruption evolves. 

We propose two approaches for this purpose, which revolve around iteratively creating interim rolling stock schedules by solving a two-stage optimiszation problem with a variety of disruption scenarios. In the second half, we focus on the integration of railway infrastructure, shunting yard and shunting driver availability into the rolling stock rescheduling framework. 

The motivation for this integration lies in the tight connection between all these problems; for an updated rolling stock schedule to be feasible, paths through the station infrastructure need to be available to transport empty rolling stock units from and to the shunting yard, a shunting plan needs to be created which feasibly parks all empty rolling stock units, and shunting driver duties that comply with labor regulations need to be created to perform the shunting movements. 

We propose a Combinatorial Bender's approach to solve this problem

About the speaker

Danny Zhu is a PhD candidate at the Econometric Institute at Erasmus University Rotterdam, supervised by Dennis Huisman and Twan Dollevoet. His research focuses on railway rolling stock (re)scheduling for Netherlands Railways (NS).

Coordinators

Ruben van Beesten